When you initially configure your agent, the content that is used is primed by the prompt you setup, but it is relying on the whole of ChatGPT (aka OpenAI). For general purpose usage, this works great, but for our Gold customers, we've had custom content models. Custom content models allow you to upload a body of content that is used by the agent to answer caller questions. This keeps the information relevant and as up to date as you like.
Starting now, you can update your own content models. We're giving everyone the opportunity to try custom content models on their agent for 30 days before asking you to upgrade to Gold. It's simple to do, as we'll show below.
What is a Model?
Agent4 organizes Custom Content by a concept called Models. You can have multiple Models, but only one can be active at any given time. You can add web sites and text files (in PDF and plain text) to this model. This allows you to experiment with different approaches until you find one that you like.
Creating your first Model
Models are managed from a web browser, so you'll visit https://app.agent4.ai and login with your normal credentials that you used to create your agent on your mobile device. Select the Content Management tab in the navigation, then select Add New Model. Give it a name and click Add.
To add content, click on Configure. This will take you to a screen that will let you add Web Sites and Text files. If you add a web site, expect a bit of a wait while we crawl your site.
You can take a look at what content we've extracted by clicking on View.
The next step is to publish your model. This will take your content and make it available for the AI agent to utilize. Click on Publish Content to do this.
Finally, you can go back to the main screen, and there will be an Interact button. Click on this, and you'll be taken to a screen where you can type a question to see how the agent would respond. The agent is not trained with these responses, this is just to test what your caller will experience.
Once you're happy with your model, all that's left to do is to click Activate and your model is live. Try it out by calling your agent from the app.
What's next?
Major improvements will be coming in the next few weeks as we roll this out.
- Prompt integration: Custom content models replace the prompt. We'll be mixing your prompt instructions with the custom content model so that you can further customize the responses (maybe you want it sound like a pirate!)
- Custom Task Models: Available today through our custom development program, custom task models allow you to introduce custom tasks that the agent can lead a caller through. Want to walk them through asking for a quote? Making a reservation? No problem. Let us show you how it works. We'll be exposing more of this functionality soon.
Need Help?
We're here! Real humans too. Send a note to [email protected] and we'll get you on track. As always, we appreciate your support and helping us grow.
